Talon SRX Motor Controller Problem

[ozrien]

Originally Posted by FTC Team CC

Ok, thanks for the clarification. Do you guys just solder the cables to each other if they run from one SRX to the next?

Also, do you have one cable on the last SRX in the daisy chain just sticking out and not connected to anything? Or does that connect back to the RoboRio/PDP?

Thanks,
Charging Champions

Hi FTC Team CC,
This is all explained in our documentation. Please read section 1.3.2 of the Talon SRX User's Guide. I think you will find answers much faster this way instead of posting a question and waiting for responses at each step.

We do not recommend star-topology, the CD forum is filled with many threads explaining why. But ultimately it can create failure-modes that are difficult to root-cause for the inexperienced. It's not clear to me that star-top is explicitly illegal from reading R70, but again we don't recommend it.

The alternative to daisy chaining would be to create a single bus harness that the CAN nodes "tap" into via short splices. This is how CAN buses are wired in cars. However in FRC its generally more convenient to daisy chain via connectors for high-serviceability during competition season.

Many teams have wired the Talon SRX on CAN bus in the fashion we've documented with great success.

There are no crash symptoms in WPILIB caused by unpowering a Talon, nor does it impact communication with sibling Talons on the CAN bus. This is explained in section 16.6 of the Talon SRX Software reference manual. In fact this is true for all of CTRE CAN products.

[FRC Team CC]

Thank you all so much for your help. We will probably just stick with CAN connections daisy-chained to each other.

Thanks,
Charging Champions

[GeeTwo]

Combining a lot of what has already been said with a bit more, here's the CAN plan we use, and why:

• CAN is designed as a bus, not a star topology. Doing a modestly deep dive, it appears that wiring in a star topology will work just fine until it fails, when it is most likely to do so with no useful diagnostic clues. Bottom line: star topology is bad juju for CAN.
• From a practical standpoint, CAN devices come in two main flavors: incorporated wires, or wire clips. (male vs female). To the extent you can, alternate the male and female devices along the bus to minimize the number of connections. This is just a specific application of KISS.
• Putting the PDP at the end of the bus is great because it has an internal terminator, but if your design has a device (e.g. the PCM on our 2016 robot) well away from the main control board, don't be afraid to use your own 120Ω resistor.
• Don't put any permanent connectors, cuts, or solder on the CAN wires. This allows you to re-use them from year to year assuming the rules don't change drastically. We had considered the new molex 2-pin genderless connectors for CAN, but when we realized how often we really needed wire ends to go in connectors on female devices, we decided not to do this - ever.
• If you have too many "male" devices compared to "female", use the VEX CAN connectors to make the necessary links. They are specifically designed for CAN impedance, and are really inexpensive - use them.
• If you have too many "female" devices compared to "male", use approx 20 AWG wire, preferably stranded, and most definitely twisted to at least "CAT 3" specs, that is a twist every 3 inches.

[FRC Team CC]

Thank you all for your help. We greatly appreciate it.

Thanks,
Charging Champions

[FRC Team CC]

One more quick question. Some of the replies on this thread have recommended using Wago Lever Nuts to connect the cables (https://www.amazon.com/Wago-222-415-...AO9N0B7PPXTIQ). Does this count as star topology (and is it therefore illegal)? Would you recommend this over the CAN connectors sold by CTR (http://www.ctr-electronics.com/adapt...-5-pack.html)? Our concern is that daisy-chaining all of the motor controllers will be more prone to failure if even one of the motor controllers in the beginning of the daisy chain fails.

Thanks,
Charging Champions

[GeeTwo]

Originally Posted by FTC Team CC

One more quick question. Some of the replies on this thread have recommended using Wago Lever Nuts to connect the cables (https://www.amazon.com/Wago-222-415-...AO9N0B7PPXTIQ). Does this count as star topology (and is it therefore illegal)? Would you recommend this over the CAN connectors sold by CTR (http://www.ctr-electronics.com/adapt...-5-pack.html)? Our concern is that daisy-chaining all of the motor controllers will be more prone to failure if even one of the motor controllers in the beginning of the daisy chain fails.

Thanks,
Charging Champions

I believe that bobbysq meant these wago lever nuts, just two connections. These will probably work OK, but the CAN connectors are more appropriately sized to CAN sized wire, making them easier to work with and less likely to reflect signals.

I understand that the motor controllers have a robust pass-through of the CAN lines; it would take a catastrophic motor controller failure to disrupt the daisy chain.

[FRC Team CC]

Originally Posted by GeeTwo

I believe that bobbysq meant these wago lever nuts, just two connections. These will probably work OK, but the CAN connectors are more appropriately sized to CAN sized wire, making them easier to work with and less likely to reflect signals.

I understand that the motor controllers have a robust pass-through of the CAN lines; it would take a catastrophic motor controller failure to disrupt the daisy chain.

Oh ok, that makes sense. Would you recommend using these at all: https://www.amazon.com/Wago-222-415-...AO9N0B7PPXTIQ? Or is it preferable to stick with the two connection options mentioned above?

Regarding the motor controllers, will the CAN connection work even if one of he motor controllers stops receiving power? Does transmission of the CAN signal rely on the motor controllers to be on? Or does the CAN signal get transmitted though the motor controllers even if it does not receive power?

Thanks once again so much.

- Charging Champions

[Ari423]

Originally Posted by FTC Team CC

Oh ok, that makes sense. Would you recommend using these at all: https://www.amazon.com/Wago-222-415-...AO9N0B7PPXTIQ? Or is it preferable to stick with the two connection options mentioned above?

Regarding the motor controllers, will the CAN connection work even if one of he motor controllers stops receiving power? Does transmission of the CAN signal rely on the motor controllers to be on? Or does the CAN signal get transmitted though the motor controllers even if it does not receive power?

Thanks once again so much.

- Charging Champions

You are better off not using the 5 connectors at all because their only use would be in a star topology which is not the way CAN was designed.

The CAN lines are passed directly through (i.e. input and outputs are soldered together). Unless the wire is cut or the speed controller catches fire and unsolders the wire, the signal will pass through even if the motor controller doesn't get power.

[Ether]

Originally Posted by Ari423

The CAN lines are passed directly through (i.e. input and outputs are soldered together).

Not necessarily disagreeing, but what is your source for this statement?

[Ari423]

Originally Posted by Ether

Not necessarily disagreeing, but what is your source for this statement?

According to page 6 of the Talon SRX User Manual, both CAN wires of the same color are "electrically identical." I would infer that to mean they are physically connected, not just connected by programming. Other people in this thread have also made this claim, but none have provided their source. I imagine someone from Vex or CTRE (or someone who has taken one apart) can give us a definite answer, but I am willing to stick with my inference from the User Manual until someone gives a more definite answer.

[bobbysq]

Originally Posted by Ether

Not necessarily disagreeing, but what is your source for this statement?

Source: Took apart a Talon SRX, looked at the wires, yep they're soldered together. Don't tell inspectors if we ever use a Talon SRX on a robot

[GeeTwo]

Originally Posted by FTC Team CC

Oh ok, that makes sense. Would you recommend using these at all: https://www.amazon.com/Wago-222-415-...AO9N0B7PPXTIQ? Or is it preferable to stick with the two connection options mentioned above?

Oh, yes we use the multiple-connector WAGO lever nuts for several power applications where we have multiple "custom circuit" loads on a single breaker, and I recommend these for that purpose -- but not for CAN bus.

[FRC Team CC]

Thank you all for the suggestions. We have ordered the CAN connectors sold by CTR electronics.

Thanks,
Charging Champions

[Hitchhiker 42]

Originally Posted by FTC Team CC

Thank you all for the suggestions. We have ordered the CAN connectors sold by CTR electronics.

Thanks,
Charging Champions

Just as an input, for our CAN connections, we use screw terminals.

[b.arci]

Originally Posted by FTC Team CC

Thank you all for the suggestions. We have ordered the CAN connectors sold by CTR electronics.

Thanks,
Charging Champions

We used Marettes (wire nuts) as a temporary connector on our competition robot this past season. Then, we soldered the CAN bus. Although, we used Marettes on our practice bot and never had a problem, and Marettes are a bunch cheaper than those connectors.